AI-Driven Document Automation for Healthcare | $12M in First-Month Revenue
The Intelligent Business Correspondence solution is an end-to-end intelligent document-processing platform that ingests, classifies and routes revenue-cycle correspondence—payments, patient statements, remittance advice and other mail—for Patient Financial Services (PFS) teams, while integrating with bank-managed lockboxes.
Project Tags
Deliverables
User Research
Visual Design
User Flows
Design Presentation
User Testing
Year
2025
Team
1 UX Designer
1 Design Manager
1 User Researcher
2 Product Managers
Duration
6 months
Role
Product Designer
IMPACT
The redesigned workflow reduced exception-handling time and gave AP teams a system that got smarter with use.
Testing and Results
Tested interactive prototype with 6 Health Information Management (HIM) professionals
100 % task completion: each user processed a document without or minimal assistance
All participants said the tool would significantly speed up their daily document-processing tasks
Personal Takeaways
Close collaboration with the dev team on this high-impact project taught me to balance bold design ideas with technical constraints, making smart compromises to deliver maximum value.
PROBLEM
Automation that couldn't earn trust left humans doing the work it was supposed to eliminate.
Why do we need to work on this?
Business Goals
Increase straight-through processing rates so invoice volume scales without headcount scaling with it.
User Goals
Stop verifying invoices the system should already be confident about.
Improve OCR accuracy over time so the product compounds in value rather than plateauing after implementation.
Reduce time-to-value for new customers by eliminating the tribal knowledge dependency that made the existing product hard to adopt.
Get enough context in the queue to make a decision without opening the invoice.
Trust that a correction made today will prevent the same error tomorrow.
User Interviews informed us Labor-Intensive Mail Prep Eats Up Hours Every Day
A researcher was brought onto the project to conduct 3 user interviews with customers using the Brainware application, as well as 1 interview with an external customer. The goal was to understand their workflows, identify what was working well, and pinpoint areas for improvement.
Manual Document Prep
“We spend more time separating and bar-coding pages than actually indexing them”
Headcount Can't Fix a System Problem
Five to six people managing one queue and still falling behind — the bottleneck was the product, not the team.
Errors Reset Every Day
Corrections were never retained, so the same vendors triggered the same failures on the same manual baseline every morning.
Journey Mapping helped Revealed Workflow Steps & IDP tool requirements
CHALLENGE
Early Design review revealed that the Design Must Leverage the Existing IDP Framework used in another solution
In the early design concepts meeting with the stakeholders, I learned due to tight resources we had to reuse an existing IDP framework, leaving limited room for customization. My task became adapting the design within those constraints while still addressing the specific needs of healthcare users. Here are a few early concepts that were appreciated but I had to move away from them
Fig. Completed Tasks did not need to be displayed
Fig. I proposed a straightforward divider line for page separation—an idea that had already earned strong user praise in a sister application.
Fig. A viewer-centric page-separation feature twas recommended to streamline implementation for the engineering team
Fig. Stakeholders appreciated the concept but asked that we extend the existing IDP framework from a sister product to minimize new-task planning
PIVOT AND FINAL DESIGN
Focusing on core tasks helped us ship a scalable IDP MVP through active engineer-designer collaboration
I adjusted my designs to prioritize essential functionality, such as document classification and metadata entry, while deferring advanced features for future updates. Additionally, I collaborated with the base IDP team to propose enhancements that would make the solution more scalable with the company’s broader needs.
Combining Page Separation and Classification in one view speeds up action-panel workflows
Automatic page separation handles most documents, but any pages the system can’t classify are flagged as issues for an admin to review and resolve.
Labeling low-confidence fields lets users quickly resolve Extraction Errors
Page Segmentation lets admins isolate and store each patient’s data when multiple patients appear within a single document
A toolbar option lets users keep or remove page sections, saving only relevant information to the selected patient’s record. When a section is removed, the rest of the page shifts up, a behavior user research preferred over the static redaction option.
Flexible QA sampling lets admins set a default review percentage and override it for specific instances
In QA Review the system rechecks classification and separation automatically—you simply confirm first-pass accuracy. In the configuration screen, you can set the percentage of documents sent to QA and override the default rate for specific categories.
Scaling the System: V2 Page Segmentation & Auto-Redaction
Why we needed to work on V2?
Fully automated segmentation lacked transparency and control, which reduced reviewer trust in high-risk healthcare workflows
Multi-patient documents increased compliance risk, requiring precise, patient-scoped validation to prevent cross-patient data exposure.
Reviewers needed automation to reduce manual redaction effort, but without sacrificing visibility, accuracy, or confidence in the output.
Flow: The system automatically identifies sections that don’t belong to the selected patient and presents them as suggested redactions. Reviewers simply accept or reject each suggestion. After every decision, the next region is automatically highlighted, creating a focused, sequential review flow that reduces context switching and speeds up validation.
Accept/Reject Controls
Undo on Hover: supporting safe and reversible actions.
Task-Level Summary Bar shows modification counts for quick progress tracking.
Patient Dropdown: Administrators can switch between patients within multi-patient documents, ensuring accurate, patient-scoped validation.
Action Toolbar with Bulk Controls: “Accept All” and “Reject All” activate once all regions are reviewed. A Preview option lets users see the final redacted document before confirming.
Page-Level Status & Controls: Displays remaining redactions per page and includes a “Remove Page” option when needed.
Manual Redaction Tools: Drawing tools allow administrators to add redaction regions if the system misses sensitive content.
Reporting Dashboards empowered teams with Insights that streamline workflows and enhance overall efficiency
Our IBC reporting dashboards deliver clear inbound and outbound summaries, track team performance metrics, and monitor document quality and number of tasks completed/still in queue. In just a glance, users gain actionable insights that streamline their workflow and enhance overall efficiency.
Next Steps
Improve the design based on customer feedback for V2.
Next Project
